Chris Wendler

Shouldn't here be a +=?

5

https://github.com/DSPsleeporg/smiles-transformer/blob/7ffa26e5bc68db0292164466398ff2edfa4a0f62/smiles_transformer/pretrain_trfm.py#L58

The TrOCR models do recognition. Thus, in order to apply it to arbitrary images, one needs a boundingbox detector. Which one would you recommend? I am currently using the one...

Recent work [1] has teased apart all kinds of labelling errors in imagenet and (rightfully) suggests that multi-label accuracy (labels for imagenet available nowadays [2]) is the better metric for...

This model does recognition. Thus, in order to apply it to arbitrary images, one needs a boundingbox detector. Which one would you recommend? I am currently using the one from...

Would be nice to have one example that implements all the tricks here.

Is it possible to call the tool from commandline with the input files?

Return the thing matched by stop_regex

2

Is it possible to access the string that has been matched by stop regex? I am working on a gsm8k tool-use example. ``` prompt = lm + ''' Solve the...

Check out our datasets, I think they might be useful for training models like this.

2

We created some large-scale multimodal datasets that contain OCR annotations, for some we ran paddle OCR over LAION images 1. https://huggingface.co/datasets/wendlerc/LAION5B-en-PaddleOCR-parquet 2. https://huggingface.co/datasets/wendlerc/LAION5B-hr-en-PaddleOCR-parquet for toand rendered images with blender, 3....

Latent downscaling

8

I have a project for which I would need latent downscaling or resizing. If you have everything set up, could you quickly train a downscaler from 128x128x4 to 64x64x4 for...

vllm backend gives different results from LanguageModel

3

 Snipped for reproducing the behavior attached. ``` # %% from nnsight.modeling.vllm import VLLM from nnsight import LanguageModel import torch llm = LanguageModel("deepseek-ai/DeepSeek-R1-Distill-Llama-8B", device_map="auto", torch_dtype=torch.bfloat16) # %% with llm.trace("this is...